XMem++: Production-level Video Segmentation From Few Annotated Frames

Hao Li, Mbzuai, Maksym Bekuzarov, Ariana Bermudez, Pinscreen

Code that ran

We lifted 1 functions out of this paper's own repositories and ran 0 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

Repository	Role	Ran
max810/XMem2	—	0 of 1

Function	Status	Where it lives
XMem	Not yet run	max810/XMem2/model/network.py pointer only (licence: GPL-3.0) · get_code("2673f38cd4224973")

Repositories linked to this paper

max810/XMem2 — marked official in Papers with Code
GCVCG/VolETA-MetaFood

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Figure 1. A demonstration of partial region segmentation (left half of the face) for extreme poses. We compare our method with the current SOTA, XMem [4], for a 1 min video (1800 frames), with only 6 frames (0.33%) annotated. No retraining or fine-tuning required.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2307.15958")
get_code_for_paper("2307.15958")
have("2307.15958")

Connect an agent — have() is free.